The Emperor's Codes Read online

  Having encoded the message into a series of five-figure groups, the operator then used an additive table book containing pages of randomly selected five-figure groups to encipher the encoded message, adding a second level of security. These additive groups were set out on each page on a 10 ¥ 10 matrix. There were two-figure indicators at the top of each column and alongside each row so that both the starting and ending points of the additive could be easily identified to the Japanese station that was to receive the message. The operator sending the message now used the Fibonacci system to add the first of the five-figure groups he had selected to the first of the five-figure groups previously encoded. Each of the additive groups in turn was added to the next code group until the entire message had been disguised from anyone who might have obtained a copy of Navy Codebook D.

  What Tiltman had done was first to identify how the system worked and, second, discover how to ‘strip away’ the cipher additive to reveal the encoded message. But although the groups for encoding numbers were known, only a few of the words and phrases represented by the five-figure code groups had been recovered.

  As for the new Japanese diplomatic cipher machine, the British machine specialists were now far too busy attempting to break the Nazi Enigma cipher. War was fast approaching and both GC&CS and FECB were preoccupied with evacuating their staffs to safer quarters.

  ‘In view of the apparent imminence of war, my particular branch of the FO received orders this morning to pick up all our secret documents etc. and to proceed to “war stations” tomorrow,’ Malcolm Kennedy recorded in his diary for 24 August 1939. The next day he drove to Bletchley Park where the Japanese diplomatic sub-section was to be accommodated in the neighbouring Elmers School.

  Left by road for war station, arriving there a bit before noon after a 60-mile run. Find that no arrangements have been made for providing us with lunch and the school which is to serve as our office building is still occupied, so we won't be able to get in our stuff and settle down to work till tomorrow at the earliest. Find too that we are to be billeted in Northampton, 23 miles from our work, so thither I took Roscoe in my car in the evening and now I am sharing a single room with him at the Angel Hotel, a 19th Century coaching inn, as accommodation is limited. Ostensibly we are now engaged in ‘Civil Air Defence’ but this cover is wearing a bit thin and why we can't admit that we are a branch of the FO heaven alone knows.

  While the FECB was not threatened by the Germans, there had been concern for some time that the Japanese might attack Hong Kong, and contingency plans had been put in place for a move to Singapore, where there was already a small group of intercept operators. During the summer, the bureau's staff had been warned that they should be packed to move at a moment's notice.

  On 25 August, the same day that the Japanese diplomatic sub-section left for Bletchley Park, the FECB staff, together with twenty lorries full of secret files, were loaded on to HMS Birmingham and sailed for Singapore. They arrived three days later, setting up their offices in Selatar barracks, in the north of the island, while the intercept operators bolstered those already in place at the Kranji wireless station seven miles away. Within two days it was in full operation, although it was some time before teleprinter links could be set up between Kranji and Selatar and messages had to be carried by motorcycle dispatch riders.

  A small team of four intercept operators, supported by one Special Intelligence officer, Squadron-Leader Alf Bennett, was left in Hong Kong. Links were also established to a number of other intercept operations on Royal Navy and Royal Australian Navy ships, at a Royal Navy site in Bombay, at a Royal Canadian Navy Base on the Pacific coast and in Melbourne.

  The FECB was given responsibility for gathering intelligence for an area stretching from the Suez Canal to the Panama Canal and for keeping track of all enemy and Allied vessels in that area, one of the bureau's officers recalled. ‘This was a very different picture from that of FECB Hong Kong, when the output had been limited to periodical summaries and a few signals.’

  The main Special Intelligence contribution to this new enlarged role came from traffic analysis, the use of routine radio communications between the units being intercepted to work out their order of battle, location and method of operation. The study of the call signs used by the stations, the subordination of each station, preambles to messages and operator ‘chatter’ provided a wealth of information about the Japanese networks the codebreakers were monitoring. This traffic analysis was aided significantly by the process of direction-finding (DF) which had been practised intermittently in Hong Kong but now became common practice whenever a new Japanese radio station appeared on the air.

  The DF stations at Kranji and Stonecutters Island were part of a network of British-controlled DF stations across the region from Canada through New Zealand and Australia to India and East Africa. Each station had an array of radio masts set up in a circular formation. From the way in which a given signal was picked up by the various aerials it was possible to produce a bearing on the enemy transmitter. By using two or more DF stations against the same target, a number of different bearings could be plotted on a map to determine the precise location of the enemy radio station.

  A few weeks after the codebreakers arrived in Singapore, Nave was married in St Andrew's Cathedral to Helena Gray, whom he had met while she was working as a nursing sister at Queen Mary's Hospital, Hong Kong. One of his close friends from GC&CS, Lieutenant-Commander Malcolm ‘Bouncer’ Burnett, who had just flown out to Singapore with the code groups Tiltman had recovered from the new General Operational Code, was the best man. Nave only had two days off for his honeymoon and shortly afterwards managed to break the new Dockyard Code.

  The bureau's vastly expanded role and the need to intercept a large depth of the new General Operational Code, in order to recover more code groups, led to increased attempts to recruit new staff, particularly Japanese linguists and intercept operators capable of taking kana Morse.

  The Japanese naval and military sections at Bletchley Park, which had each contained only four codebreakers when the European war broke out, were slimmed down still further to find experienced officers for Singapore. By May 1940 the penetration of the codebook used as the basis for the main General Operational Code had progressed to the stage where stereotyped messages such as convoy schedules and individual ship movements could be decoded, although it was still not possible to read detailed operational orders.

  Extra intercept operators were obtained by retraining Royal Malayan Navy signallers. But demands made on manpower by the outbreak of war in Europe meant that finding good intercept operators was difficult and as a result the navy began training Wrens to take kana Morse so they could be sent out to Singapore. They were trained as wireless telegraphists at King's College, Campden Hill Road, Kensington, in west London. Joan Sprinks, from Norwich, was twenty-three when she joined the course.

  It was expected to take six months and the weekly payment was to be 33/6d [£1.67], of which £1 a week was to be deducted for board and lodging. I had no experience at all at this sort of thing but some of those on the earlier courses were already trained telegraphists and two had actually served in the same job during the First World War. Life at the depot was a mixture of Morse, drill, PT, assorted lectures, and ‘darkening the ship’ for the blackout.

  The need for intercept operators was now so urgent that the kana Morse course was cut down from six months to three. The Wrens were then sent out to one of the main Royal Navy intercept sites at Flowerdown, near Winchester in Hampshire, or Scarborough in Yorkshire. Joan Sprinks was sent to Scarborough.

  We were given brass buttons and rated as Chief Wrens so as to be equal in status to the civilian men's trade union. At Scarborough we were billeted in boarding houses and walked or cycled to go on watch in a room that was below ground. The night watch took a lot of getting used to – from 11.30 p.m. to 8 a.m. sitting at a bench with a number of ex-naval personnel who all seemed to consume very strong tea and to smoke endle
ss ‘ticklers’, which were hand-rolled with very strong navy-issue tobacco. There was no fresh air at all, although the watchroom was occupied for the whole of the twenty-four hours and in addition there was of course a total blackout. Everything was so secret that we were trusted not to breathe a word about the work we were doing, which was intercepting the messages of German ships.

  A few months later Sprinks and her fellow Wren wireless intercept operators were asked if they were prepared to serve overseas.

  We learned later that this was to be in Singapore – at the wireless telegraphy station at Kranji. We reported to the Royal Naval College, Greenwich, for a special course under civilian instructors and met the officer who was to be in charge of us – Second Officer Betty Archdale, who had been a barrister and captain of the Women's Cricket Team to tour Australia in the 1930s.

  Kirk Gill was one of the service instructors allocated to teaching kana Morse to new recruits. An RAF sergeant instructor, he was well aware of the difficulties they faced, having been thrown in at the deep end himself.

  Air Ministry decided that we should concentrate on teaching Japanese kana Morse. It did not matter that we hadn't any idea what it was all about. Just do your duty and the devil take the hindmost. Frantic swotting was the order of the day to combat this mammoth task. We not only had to memorize the code but to be able to transpose it into Japanese language which was a very formidable task.

  The kana Morse Code had seventy-five different characters. It employed a mixture of the standard Morse symbols to which the wireless operators were accustomed plus a number of ‘barred’ letters, combinations of two standard symbols run together. But neither the standard symbols nor the barred combinations of letters represented the same thing in kana Morse as they did in the international code. The standard Morse code symbol –…, for example, which is B in the international code, was HA in the Japanese code. A dash on its own, T in the international code, was MU in the kana code, while the barred letter BT, ie –…–, was ME in the kana code. In addition, the operators had to cope with a system of suffixes, Hanigori or Nigori, designed to provide an inflection to the character: Hanigori, sent as..–.–, denoting a soft sound, and Nigori, sent as.., denoting a hard sound.

  The British intercept operators took the characters down Japanese style, in columns, rather than working left to right. But in order to ease the process of learning the Japanese system, they wrote them initially as if they were standard international characters and then transcribed them into the kana versions afterwards using a table.

  Having mastered the complicated Japanese Morse code, the new Wren operators set off for Singapore, leaving from Greenock in Scotland. They sailed on board the Blue Funnel ship SS Nestor in a convoy of about fifty ships protected by the aircraft carrier HMS Argus and its escort vessels.

  ‘A few days out of port a lone Focke-Wulf Kondor attacked the convoy, dropping several bombs without apparently scoring a direct hit,’ said Joan Sprinks. ‘But they were close enough for two of us to finish a whole box of Black Magic chocolates, very precious in those days, just in case we, and they, went to the bottom.’

  6

  PURPLE MAGIC

  The increasing use of the Japanese Type B machine had restricted the amount of high-grade Japanese enciphered diplomatic traffic the British were breaking. But there had been enough indications in the Type A traffic of what was to come to ensure they were not caught out by the signing in September 1940 of the Tripartite Pact, binding Germany, Italy and Japan together in a full-blooded military alliance.

  The three Axis powers agreed to recognize each other's expansionist claims in Europe, Africa and Asia. More importantly, they promised to come to each other's assistance in the event of military intervention in either Europe or South-east Asia by any power not already involved, a transparent allusion to the USA.

  Malcolm Kennedy was dismayed by Japan's decision to sign the pact. ‘Though the announcement of this alliance has not come as a complete surprise to those in the know it is nonetheless unwelcome,’ he noted in his diary. ‘One cannot but feel that Japan herself will live to regret what she has done as, compared with Germany and Italy, she has precious little to gain by it and the Dickens of a lot to lose.’

  Although the Type A machine would not be totally replaced for another year, Kennedy's diplomatic section was struggling. It managed to intercept messages ordering all Japanese missions abroad to co-operate with their Italian and German colleagues in collecting intelligence on the movement of British and American warships, aircraft and troops because the message circulation was so large that the Type A machine had to be used. But the main Japanese diplomatic cables between Berlin, Rome, London, Washington and Tokyo were now being sent on the Type B machine and frequently the latest Japanese moves were clearly as much ‘a complete surprise’ to Kennedy as they were to anyone else.

  As the Japanese diplomatic section had found to its cost, the bombing of British towns was in full flow. On 21 November 1940 Kennedy had arrived at the office in Elmers School to find that it had been bombed overnight.

  Typists’ room and telephone exchange blown to bits by a direct hit and the Vicarage next door damaged by another bomb which landed in the garden. A third bomb exploded in the road outside, while two more landed over at the Park, one of them bursting a bare half-dozen paces from Hut 4. By great good fortune there were no casualties. We however have been moved to the room used by the South American section, who in turn have been transferred to the Park.

  The decision to move Kennedy's section is a reflection of the low priority accorded to Japanese material at Bletchley Park at the time. Its research section was completely preoccupied with breaking German machine ciphers. With their backs very firmly against the wall, the British could only concentrate on one problem at a time and the battle to break Enigma took first place. All over Bletchley Park, prefabricated huts were being put up, but none of them was to work on Japanese radio traffic.

  That was considered the sole prerogative of the Far East Combined Bureau. It was concentrating the bulk of its resources on the Japanese Navy General Operational Code, which made very heavy demands on manpower. The Type B machine was simply not regarded as a high enough priority to throw resources at it. The British codebreakers in Singapore did at one point come close to obtaining a ‘pinch’ of a Type B machine, Harry Shaw recalled. But they were prevented from doing so by bumbling bureaucracy.

  A Tokyo cipher message gave the date of arrival, and name of ship, in which Cipher Machine B was being sent to the Japanese Consulate in Singapore under diplomatic immunity. The police were approached to obtain their connivance in engineering an ‘accident’ while the cases containing the machine were being hoisted out of the ship, enabling a technical officer to view the contents. The police refused to co-operate without the tacit support of the Colonial Secretary, Singapore, who strongly opposed the scheme, and the project was dropped.

  If the British had decided to place the breaking of the Type B traffic on the back burner, the Americans had not. The US Army codebreaking organization, the Signal Intelligence Service, had been concentrating on it since it first appeared in late February 1939. On 27 September 1940, the day the Tripartite Pact was signed, they succeeded in making the first break into the Type B machine.

  The new machine was also known as the 97-shiki oo-bun inji-ki, Alphabetical Typewriter 97, indicating that it was first developed in 1937, the Japanese year 2597. Like the Type A machine, it was electro-mechanical and had two typewriter keyboards and a plug-board. There were, however, a number of fundamental differences between the two machines. The principal difference was that the encipherment circuits on the Type B machine ran not through a rotor, as on its predecessor, but through a series of telephone stepping switches of the same type as those used by Harold Kenworthy to produce the J machine with which the British were deciphering the Type A traffic.

  A single main switch controlled the encryption of six letters which, unlike the Type A machine, were
not the vowels but changed daily. The other switches were organized in banks of three and controlled the remaining twenty letters. The switches were designed to simulate the action of the rotors on a standard cipher machine such as the German Enigma machine, moving to change the method of encryption as each letter is typed in. The single main switch stepped one level every time a letter was keyed in. The other three banks moved at different speeds, much like the rotors on an Enigma machine. The first bank of switches was the ‘fast’ bank, stepping once for the first twenty-four key strokes. When the twenty-fifth character was typed in, the fast bank stayed where it was and the second ‘medium’ bank of switches stepped once. The second bank also only moved twenty-four times before the third or ‘slow’ bank came into play. At the end of the next full ‘rotation’ of the fast bank of switches, the 625th operation of the machine, both the fast and medium banks remained where they were while the slow bank stepped once.

  The result was a much more complex encipherment process where the substitution produced at each point was completely unrelated to any of the others, unlike a rotor system where all the substitutions produced by the wheel follow each other in rotation.

  Although this appeared to make the Type B machine far superior to the German Enigma machine, it was paradoxically much less secure. The Enigma had a number of differently wired rotors which could be changed and shuffled through a wide variety of permutations. But the Type B's stepping switches were all fixed into the machine, drastically cutting the number of encipherment options. The real difficulties generated by the machine lay in the tangled web of wiring connecting the banks of stepping switches. If any potential codebreaker were to work out the wiring of the Type B machine, the enciphered traffic would be broken with ease.